How to stop fearing AI and increase profit with technology

This fear was born out of past experience with IT rollouts.

How many such stories have we (and you) heard: at first the company expected to invest two million and launch an MVP in three months.

But a year and a half passes, the total bill already exceeds 15 million, and the system still has not been launched.

Endless development is such a common problem that our on-time IT projects often surprise clients. AI also carries the risk of getting stuck in endless implementation with a ballooning budget.

Add to that the unpredictable payoff from integration, and it becomes clear why businesses don't want to take the risk.

It makes it easier to find agreements through a chatbot.

You and your staff no longer have to reread hundreds of emails and chats to find the information you need.

Just ask the AI assistant a question and get an answer in 15 seconds.

It analyzes the call against important parameters: setting the agenda, adherence to the script, the manager's politeness, willingness to solve the client's problems, recording agreements, rapport markers, and so on.

It automatically emails the summary to you and the client, and updates the client record in the CRM and the project documentation.

Imagine you have an AI assistant for tenders.

Based on the set criteria, it assesses whether the company should take part in the selection.

You are confident that the AI selects all relevant offers and submits bids for them.

But one day, while waiting in line at a cafe, you meet a friend and learn that his company held a tender that would have been a good fit for you, and you ignored it.

This is an alarming signal: if the neural network got this wrong, how many other mistakes did it make? An AI assistant is like an employee who knows a great deal but is still inexperienced.

At the start of its work, it has a poor "understanding" of what is right and wrong.

This is a critical and costly stage that determines whether the AI will work flawlessly. The specialist who has always handled your tenders will need to validate every decision the AI assistant makes and confirm or reject it.

This approach helps the AI accumulate experience (a dataset) and avoid mistakes in the future.

If the work requires responsible decisions or a creative approach and a human is indispensable, delegate only part of the tasks to the virtual assistant. For example, an AI assistant can take on the estimation of a tender's technical specification: it breaks large tasks into smaller ones according to an approved pattern and calculates each stage in hours and money. And if the assistant encounters a non-standard request, it forwards it to a human.

This will save the tender specialist time, since they will only need to handle the tasks where a decision cannot be made without them.

On top of that, the number of tender bids will definitely grow.

Tomorrow the entrepreneur decides to also take part in design tenders. It seems that all the algorithms will have to be rewritten from scratch and checked to make sure everything works correctly. So does that mean investing another couple of million in a new AI assistant? No, there is no need to implement a new AI assistant.

The target orders and the method for evaluating requirements have changed, but most of the automated steps remain the same. For example, tenders still need to be gathered from the same platforms, and they still need to be analyzed for relevance and the chance of winning, including whether there are signs of a rigged tender.

Regardless of the technical specification, you need to collect information from the required documents.

There is no need to retrain the assistant for these stages.

But some refinement and additional training will still be needed: you will have to add new keywords to select tenders of interest.

A new method for evaluating requirements will need to be developed, since the list of tasks in a design tender differs from a construction order.

But the additional training process is many times faster and cheaper than implementing a new assistant: you simply load the up-to-date guidelines and files into the AI system.

Then these procedures will be automatically uploaded to your assistant, after which you run tests and everything is ready.

At our company, AI is used by HR specialists, project managers and developers.

Here is what we saw after KT.Team employees started using AI assistants. HR.

We automated pulse surveys: the HR assistant processes survey data and highlights pain points, growth opportunities, and work issues for each employee. HR specialists focused on addressing these pain points, which helped us improve the employee loyalty index (eNPS).

Project manager. The AI assistant transcribes all calls word for word and stores them.

When a misunderstanding arises on a project or previous agreements need to be revisited, the project manager can ask the AI assistant about them within a few minutes and get a detailed answer with a reference to the source, date and time.

The manager can focus on communicating with the team and clients instead of spending hours re-listening to old meetings. Developer.

The key part of a developer's work is understanding the task correctly: what value the feature should deliver and what outcome the client wants to achieve.